Martinez v. Delta Airlines, Inc. does not exist — a fabricated AI citation

No Westlaw entry exists at 2019 WL 4413263, and no Texas Court of Appeals panel decided a case captioned “Martinez v. Delta Airlines, Inc.” in September 2019 or at any other time matching this citation. It is the fourth of the six fabricated authorities ChatGPT produced for the now-infamous Mata v. Avianca opposition brief.

A Westlaw-style “WL” citation is designed to look unverifiable-by-casual-reading — no reporter volume to flip to, just a database number — which arguably makes it one of the more effective fabrication formats an AI system can generate, since a reader cannot spot-check it against a physical reporter the way they could a traditional F.3d or So.2d citation. It still fails the only test that matters: it does not resolve to a real document in Westlaw, Lexis, or CourtListener.

Judge Castel's sanctions opinion treated all six fabricated cases, including this one, as a single pattern of conduct rather than isolated errors, because the attorneys never opened or verified any of them before submission — the deciding factor in the court's finding of bad faith under Rule 11. The lesson generalizes well beyond aviation law: any citation format that resists casual spot-checking — a WL number, an unpublished order, a slip opinion — deserves more verification effort, not less, precisely because it is harder to verify by eye.